Re: [PATCH 1/3] dma mapping benchmark: modify the framework to adapt to more map modes

[Barry Song]

On Wed, Feb 12, 2025 at 3:27 PM Qinxin Xia <xiaqinxin@xxxxxxxxxx> wrote:
>
> In this patch map_benchmark abstract in four interface: prepare, unprepare,
> do_map, do_unmap. When there's a new mode to add, need four steps:
> 1) Add the mode in map_benchmark.h
>
> 2) Defines the mode param, like struct dma_xxx_map_param, and this object
>    will be return in prepare and as input parameter in other ops;
>
> 3) Defines the ops functions:prepare, unprepare, do_map, do_unmap.
>
> 4) Add the new mode in dma_map_benchmark_ops.
>
> Signed-off-by: Qinxin Xia <xiaqinxin@xxxxxxxxxx>
> ---
>  include/linux/map_benchmark.h |   6 ++
>  kernel/dma/map_benchmark.c    | 122 +++++++++++++++++++++++++++-------
>  2 files changed, 105 insertions(+), 23 deletions(-)
>
> diff --git a/include/linux/map_benchmark.h b/include/linux/map_benchmark.h
> index 62674c83bde4..054db02a03a7 100644
> --- a/include/linux/map_benchmark.h
> +++ b/include/linux/map_benchmark.h
> @@ -15,6 +15,11 @@
>  #define DMA_MAP_TO_DEVICE       1
>  #define DMA_MAP_FROM_DEVICE     2
>
> +enum {
> +       DMA_MAP_SINGLE_MODE,
> +       DMA_MAP_MODE_MAX
> +};
> +
>  struct map_benchmark {
>         __u64 avg_map_100ns; /* average map latency in 100ns */
>         __u64 map_stddev; /* standard deviation of map latency */
> @@ -27,5 +32,6 @@ struct map_benchmark {
>         __u32 dma_dir; /* DMA data direction */
>         __u32 dma_trans_ns; /* time for DMA transmission in ns */
>         __u32 granule;  /* how many PAGE_SIZE will do map/unmap once a time */
> +       __u8  map_mode; /* the mode of dma map */

We previously added some padding to ensure the uABI remained consistent.
I just noticed that Tiantao’s commit (8ddde07a3d285a0f3cec, "dma-mapping:
benchmark: extract a common header file for map_benchmark definition")
accidentally removed that padding, which has completely broken the ABIs.

Could you send a patch to fix this regression, and CC it to
stable@xxxxxxxxxxxxxxx before adding the new field which should
use the expansion[] instead.


>  };
>  #endif /* _KERNEL_DMA_BENCHMARK_H */
> diff --git a/kernel/dma/map_benchmark.c b/kernel/dma/map_benchmark.c
> index cc19a3efea89..d8ec0ce058d8 100644
> --- a/kernel/dma/map_benchmark.c
> +++ b/kernel/dma/map_benchmark.c
> @@ -5,6 +5,7 @@
>
>  #define pr_fmt(fmt)    KBUILD_MODNAME ": " fmt
>
> +#include <linux/cleanup.h>
>  #include <linux/debugfs.h>
>  #include <linux/delay.h>
>  #include <linux/device.h>
> @@ -31,17 +32,98 @@ struct map_benchmark_data {
>         atomic64_t loops;
>  };
>
> +struct map_benchmark_ops {
> +       void *(*prepare)(struct map_benchmark_data *map);
> +       void (*unprepare)(void *arg);
> +       int (*do_map)(void *arg);
> +       int (*do_unmap)(void *arg);
> +};
> +
> +struct dma_single_map_param {
> +       struct device *dev;
> +       dma_addr_t addr;
> +       void *xbuf;
> +       u32 npages;
> +       u32 dma_dir;
> +};
> +
> +static void *dma_single_map_benchmark_prepare(struct map_benchmark_data *map)
> +{
> +       struct dma_single_map_param *mparam __free(kfree) = kzalloc(sizeof(*mparam),
> +                                                                   GFP_KERNEL);
> +       if (!mparam)
> +               return NULL;
> +
> +       mparam->npages = map->bparam.granule;
> +       mparam->dma_dir = map->bparam.dma_dir;
> +       mparam->dev = map->dev;
> +       mparam->xbuf = alloc_pages_exact(mparam->npages * PAGE_SIZE, GFP_KERNEL);
> +       if (!mparam->xbuf)
> +               return NULL;
> +
> +       /*
> +        * for a non-coherent device, if we don't stain them in the
> +        * cache, this will give an underestimate of the real-world
> +        * overhead of BIDIRECTIONAL or TO_DEVICE mappings;
> +        * 66 means evertything goes well! 66 is lucky.
> +        */
> +       if (mparam->dma_dir != DMA_FROM_DEVICE)
> +               memset(mparam->xbuf, 0x66, mparam->npages * PAGE_SIZE);
> +
> +       return_ptr(mparam);
> +}
> +
> +static void dma_single_map_benchmark_unprepare(void *arg)
> +{
> +       struct dma_single_map_param *mparam = arg;
> +
> +       free_pages_exact(mparam->xbuf, mparam->npages * PAGE_SIZE);
> +       kfree(mparam);
> +}
> +
> +static int dma_single_map_benchmark_do_map(void *arg)
> +{
> +       struct dma_single_map_param *mparam = arg;
> +
> +       mparam->addr = dma_map_single(mparam->dev, mparam->xbuf,
> +                                     mparam->npages * PAGE_SIZE, mparam->dma_dir);
> +       if (unlikely(dma_mapping_error(mparam->dev, mparam->addr))) {
> +               pr_err("dma_map_single failed on %s\n", dev_name(mparam->dev));
> +               return -ENOMEM;
> +       }
> +
> +       return 0;
> +}
> +
> +static int dma_single_map_benchmark_do_unmap(void *arg)
> +{
> +       struct dma_single_map_param *mparam = arg;
> +
> +       dma_unmap_single(mparam->dev, mparam->addr,
> +                        mparam->npages * PAGE_SIZE, mparam->dma_dir);
> +       return 0;
> +}
> +
> +static struct map_benchmark_ops dma_single_map_benchmark_ops = {
> +       .prepare = dma_single_map_benchmark_prepare,
> +       .unprepare = dma_single_map_benchmark_unprepare,
> +       .do_map = dma_single_map_benchmark_do_map,
> +       .do_unmap = dma_single_map_benchmark_do_unmap,
> +};
> +
> +static struct map_benchmark_ops *dma_map_benchmark_ops[DMA_MAP_MODE_MAX] = {
> +       [DMA_MAP_SINGLE_MODE] = &dma_single_map_benchmark_ops,
> +};
> +
>  static int map_benchmark_thread(void *data)
>  {
> -       void *buf;
> -       dma_addr_t dma_addr;
>         struct map_benchmark_data *map = data;
> -       int npages = map->bparam.granule;
> -       u64 size = npages * PAGE_SIZE;
> +       __u8 map_mode = map->bparam.map_mode;
>         int ret = 0;
>
> -       buf = alloc_pages_exact(size, GFP_KERNEL);
> -       if (!buf)
> +       void *arg = dma_map_benchmark_ops[map_mode]->prepare(map);
> +
> +       if (!arg)
>                 return -ENOMEM;
>
>         while (!kthread_should_stop())  {
> @@ -49,23 +131,10 @@ static int map_benchmark_thread(void *data)
>                 ktime_t map_stime, map_etime, unmap_stime, unmap_etime;
>                 ktime_t map_delta, unmap_delta;
>
> -               /*
> -                * for a non-coherent device, if we don't stain them in the
> -                * cache, this will give an underestimate of the real-world
> -                * overhead of BIDIRECTIONAL or TO_DEVICE mappings;
> -                * 66 means evertything goes well! 66 is lucky.
> -                */
> -               if (map->dir != DMA_FROM_DEVICE)
> -                       memset(buf, 0x66, size);
> -
>                 map_stime = ktime_get();
> -               dma_addr = dma_map_single(map->dev, buf, size, map->dir);
> -               if (unlikely(dma_mapping_error(map->dev, dma_addr))) {
> -                       pr_err("dma_map_single failed on %s\n",
> -                               dev_name(map->dev));
> -                       ret = -ENOMEM;
> +               ret = dma_map_benchmark_ops[map_mode]->do_map(arg);
> +               if (ret)
>                         goto out;
> -               }
>                 map_etime = ktime_get();
>                 map_delta = ktime_sub(map_etime, map_stime);
>
> @@ -73,7 +142,9 @@ static int map_benchmark_thread(void *data)
>                 ndelay(map->bparam.dma_trans_ns);
>
>                 unmap_stime = ktime_get();
> -               dma_unmap_single(map->dev, dma_addr, size, map->dir);
> +               ret = dma_map_benchmark_ops[map_mode]->do_unmap(arg);
> +               if (ret)
> +                       goto out;
>                 unmap_etime = ktime_get();
>                 unmap_delta = ktime_sub(unmap_etime, unmap_stime);
>
> @@ -108,7 +179,7 @@ static int map_benchmark_thread(void *data)
>         }
>
>  out:
> -       free_pages_exact(buf, size);
> +       dma_map_benchmark_ops[map_mode]->unprepare(arg);
>         return ret;
>  }
>
> @@ -209,6 +280,11 @@ static long map_benchmark_ioctl(struct file *file, unsigned int cmd,
>
>         switch (cmd) {
>         case DMA_MAP_BENCHMARK:
> +               if (map->bparam.map_mode >= DMA_MAP_MODE_MAX) {
> +                       pr_err("invalid map mode\n");
> +                       return -EINVAL;
> +               }
> +
>                 if (map->bparam.threads == 0 ||
>                     map->bparam.threads > DMA_MAP_MAX_THREADS) {
>                         pr_err("invalid thread number\n");
> --
> 2.33.0
>

Thanks
Barry